Daily life in the lab revolves around doing experiments. The job is (1) to expand the boundaries of human knowledge and (2) communicate your findings across the field, and potentially beyond it. Whether you’re an undergraduate just starting on your first research project, or a seasoned PI, some element of these two tasks will make up your career. As you get higher in the science ranks, you might add (3): training the next generation of scientists.
Upon first joining a lab, you’ll hopefully be given a project, or join a team all working on the same project. The first thing you’ll spend considerable time doing is reading about it (See: On reading papers). But you’ll also spend a lot of time learning the various bits of the trade for the lab; where the tubes are kept, where’s your freezer space, etc. It will be a little mysterious, and especially if you’ve never worked in a lab before, it will be daunting.
First word of advice when you first join a lab—write everything down. One of my pet peeves as a mentor is having a mentee of any stage nod along and then an hour later, ask a question that I already answered. I get it, we all forget things and make mistakes, but sometimes, it could be avoided with a little forethought and not assuming you’ll memorize everything with one pass—because let’s be honest, that rarely happens. When you first join a lab, have a spiral bound notebook in addition to your lab notebook and be ready to write in it at all times. I hope someone gives you your official lab notebook—they should. Too many times I’ve had to buy my own, and on giving them (back?) to the university, I’ve missed the notebooks terribly as my own works of art.
I’m sorry, but you cannot expect to acquire good data right away. Nor will you master the techniques right away. Does it happen sometimes, to some graduate students, yes. Should you count on it, hell no. It would be a disservice to deny yourself the great privilege of repetitive failure and frustration. Because failure and frustration are inevitable in the sciences, by early and often feeling failure, figuring things out, and then moving forward, you give yourself the incredible gift of knowing what failure feels like, and the knowledge that you can move forward from it. “I’ve been here, but I got through it once, and I can envision that sooner or later, I’m going to get through it again.”
Considerable time will be spent learning your trade, screwing up the techniques, analyzing the wrong things, optimizing assays, figuring out what it is you can measure, and where the valuable pieces of information you can mine out of your project lie. In graduate school, it took two years for me to have real confidence as an electrophysiologist, and that wasn’t nearly the end of my learning. Of course, this will depend on how “pre-optimized” your inherited techniques. If you’re using fairly standard, established techniques that the lab has already published, you can probably hope to be collecting useful data within a year or two as a graduate student, about six months to a year for a new post doc. If you’re breaking new ground where there is no standard set of rules, hence, you’re responsible for making rules that are both rigorous and yield interesting data, give yourself two-three years as a graduate student, and one to two years as a new postdoc. I’m estimating the graduate student time frame from the date you join a lab, not from the date you entered graduate school (you typically will have a year of rotations). And these numbers are really just some off-the-cuff estimates, that can be influenced by a myriad of circumstances. The main takeaway is to be patient and spend time building confidence in your technique and your field of study.
That said, let’s talk for a while about how you get these rules and feel confident in them. Whether you are just embarking on a project, or are in the midst of collecting data for a publication, allow me to share a bit of wisdom I’ve gained over the years. (Note, that since most of my training has been with individual live cells, and I have close to no experience with “big data,” these bits of advice apply mostly to “smaller” sorts of experiments. Ignore as you see fit, though there might be a gem or two that resonates:)
To do good science, repetition is your best, most honorable, most cherished, friend. Repetition, repetition, repetition. On the contrary, perfection is your enemy. Reading papers and dreaming of experiments is fun, but unless you’re writing a grant/paper/review, that is not the momentum that drives you forward. Trying to earn the praise and well-wishes of your PI is a minatory sludge monster that will haunt you when you try to sleep at night. In graduate school, the bulk of your time should be spent collecting and analyzing data— then let the conclusions from those data take you to the next experiment. Do not seek praise. Seek data. To get data, you need repetition.
Repetition really means doing the same thing over again. Get an n of three (at the very least!) before making any conclusion. The average wants to help you. It wants you to succeed and know the truth. But you cannot make an average out of an n = 1 or 2. Do not spend tons of energy focusing on what happened during one particular trial. Keep your ass in that chair and do the experiment two more times! These are complex systems, and there is no way you can predict every bit of nuance that’s going to happen. The cell starts to run down? That’s okay. Throw it in the average. An outlier test will be able to tell you later if it really needs to be scrapped. The cell starts to drift out of frame. Can you still analyze it? Great. Throw it in the average. The key in my book is doing the exact same thing as many times as feasible, and then moving on to another exact same thing as many times as feasible (in electrophysiology, the magic number is often 6 cells). If “as many times as possible” is too much, a power analysis should help you know how many you should do.
Of course, make a note if you know something isn’t right. The coverslip is leaking, the pump just stopped working, the microscope exploded, I imaged the wrong cell, the slide just cracked and destroyed the objective, the mouse was just possibly raptured and we can’t find it —these are all worthy reasons to either call quits on an experiment or at least mark it as “don’t waste time analyzing this.” But do not tempt the damning thought: “I don’t think this experiment is going the way it should, so I should stop it.” That is to pose your own judgement on the experiment, and the cells/mouse/fish/neuron/whatever will have their vengeance. They must speak for themselves, regardless of your hypothesis. Think instead: the sooner you have an average of something, the sooner you have something to call “preliminary data.” It won’t go in a paper, maybe, but it might look stellar in a grant with a power analysis beside it. Go the extra mile, get an average.
In this regard, don’t be afraid to push yourself now and then. When your brain goes “I’m tired,” or “I’m hungry, I just want to go home,” or the bane of science, “it’s not perfect, so I give up”: This is the time for discipline. There is a difference between self-discipline and self-flagellation. One is crucial for success. The other is the antithesis of it.
Repetition is necessary for learning techniques too, if you can feasibly repeat them. If there’s a part of an experiment that is tricky and you can practice with some left over slices, cells, fish, etc, make time for practice. In my early postdoc, I reserved an afternoon for practicing how to mount fish for microscopy, with no intention of actually imaging those fish. As with all learning, expect to make mistakes. When you make them, be aware, and try to correct. I believe the most fundamental part of learning anything is self-awareness; knowing where you’re struggling and where you could use some help, and also what you do well. Early in my career (both in writing and in science)(careers?) I used to think it was someone else’s job to tell me my strengths and weaknesses. You can want that, but it’s not likely to happen, and other’s perceptions are probably not correct. Take it as your job. Of course, some things, like a surgery, you’re not going to be able to “just practice.” On those, understand that you can only improve over the long term and if you keep coming back to it.
Long experiments are a hallmark of the sciences. You might have to babysit whatever it is you’re experimenting for five minutes, fifteen minutes, an hour, two hours. What do you do in these time windows? The early graduate student will try to read papers. I know—I’ve done it. It doesn’t work well. Unless all you need is a superficial knowledge of the paper, don’t bother, it is too much to have one half of your brain doing science and the other half of your brain trying to understand someone else’s science. I have found that the best things you can do during long experiments are things that help you stay there and stay happy: at the scope, rig, whatever devise you’re using, most likely in a dark room. Being happy is the key to staying some of those long hours to get multiple experiments done. In graduate school, for me, that was often getting writing in. One way to get me through 4-5 experiments of 1.5 hr each, beginning around 5pm until after 10pm (this was after a full 8 hour lab day, mind, so I wasn’t exactly bright and chipper and fresh coming to this) was having a printed copy of my novel in front of me, a marking pen, and the Lord of the Rings soundtrack. I would stay at the scope all night for that! (Good memories). I can remember critical scenes of my novel written “at the rig” or “at the confocal.” I know people who have watched tv shows, Skyped with their parents and family, watched live soccer and screamed at the computer screen, choreographed dance routines—whatever it takes. If you can set up an experiment to run all night and automatically sample for you, do that. If you can’t, enjoying where you are is critical.
Once you have done the experiment (more than once), the next step is to analyze the data. Set aside a significant portion of time to do this. When I was a graduate student, a mentor once told me to spend at least as much time analyzing the data as running the experiment. Of course, once you establish a work flow, analyzing data can be quick, but you want to be on the lookout for abnormalities, and you want to be careful to not make mistakes, because it can be so easy to make mistakes. Don’t analyze data hungry or tired. Be cautious in your analysis. Never sell your soul on a conclusion.
I am a happy person if I can trust the analysis. The more “handwaving” that gets introduced while analyzing (“Well, it works if I do this…but I just have to stick to that parameter and that’s it…”) (“If I just have a bit of this other cell in the frame it won’t matter too much, right?”), the less happy I am. Happy data are trusted. Unhappy data are more than not scrapped. You can’t waste your science time scrapping data, life is too short and you have other things on which to focus your scientific career. Happy data emerge from doing multiple experiments (n > 3) and analyzing them well, so that they mean something solid and trusted that you could summarize to a non-scientist in one sentence. (“X does this to Y.” “P is reduced from Q.”)
Running experiments and analyzing them, making them into an understandable truth, is the heart of the scientific job. You should count on presenting your results at a lab meeting, departmental meetings, and a few times a year, conferences. My graduate school lab had a general rule of “produce one figure a week.” At whatever pace is appropriate, you should be regularly finishing experiments, analyzing them, and running new ones whether the timeline is on a week scale, a month scale, or a multi-month scale. Always let the data guide you to the next set of experiments, not your own expectations.
Andrea L McQuate 